Alibaba Cloud's service team went deep into Krypton's business, conducted a comprehensive survey and assessment of the customer's APP status based on the eight SRE systems, identified technical architecture problems and risks, and exported an overall transformation plan for Krypton's original APP cloud architecture, focusing on 14 special projects, including full link grayscale publishing, microservice governance, capacity planning, elastic scaling, and cloud product optimization.

As the sales of Krypton cars become more and more popular, their registered users and daily active users grow rapidly, and more and more business scenarios and new functions need to be supported, with an average upgrade frequency of a small version every two or three days and a large version every half a month. Under the original architecture, in order not to affect the daytime business peak, each version can only be released in the early morning business low peak period. Alibaba Cloud service team helps customers develop and launch canary release plans: deploy grayscale versions and perform grayscale verification according to traffic ratio. After verification, release the production environment and cut the stream, so as to meet the demand of releasing polar krypton small versions at any time during the day. For the scenario where multiple microservices on the polar Krypton core business link need to be issued at the same time, the full link grayscale of multiple services is achieved based on MSE cloud native gateway and traffic grayscale marking, covering grayscale scenarios such as CDN, gateway, MQ, configuration, and database. Multi service daytime release is realized without changing any business code. At the same time, it is verified through gradual traffic amplification. If there is a problem, the traffic can be cut back in time, reducing the stability risk that may be caused by daytime release. Through the transformation of cloud efficiency pipeline, it helps Krypton realize the automatic release of core business and better improve the deployment efficiency.

The Krypton APP application uses a microservice architecture. When doing business version publishing, elastic capacity expansion and other scenarios, it will encounter problems such as increased request failure rate and constant restart of POD. In view of this problem, combined with the product capabilities of the microservice engine MSE, we realized the lossless online and offline publishing of microservices by means of adaptive waiting and active notification during the application offline process, readiness inspection during the application online process, service preheating, etc., effectively avoiding the loss of flow during the publishing process, and reducing the risk of business access failure. The MSE traffic prevention and control capability is introduced, and the corresponding technical means are implemented for the core business scenarios, such as interface flow restriction and degradation, MQ peak shaving and valley filling, slow SQL flow restriction management of database, etc., to improve the overall stability of the service.

With the rapid development of business, the problem of insufficient capacity under the original architecture of Krypton APP has become increasingly prominent. In the face of new car release, sales activities, and sudden hot spots, it is impossible to rapidly expand horizontally. In addition, a large number of core business databases are placed on the same database instance, which is prone to "damage and loss". Alibaba Cloud recommends PolarDB-X products to Krypton Polar, and cooperates with Krypton Polar to separate the core databases such as the point database and user database one by one. It solves the problem of too large a single table through horizontal splitting of large tables, and improves the horizontal elastic capacity expansion capability of the database layer. In addition, for the pain point of insufficient elasticity of microservices, output container elasticity schemes such as elastic expansion of multi zone nodes, HPA, CronHPA, etc., to improve the core service's ability to respond to traffic emergencies.

Krypton APP uses more AliCloud products, and its business has developed rapidly. There are many unreasonable ways to use Krypton APP. Through multiple rounds of scheme discussions, Alibaba Cloud has formulated a cloud native product optimization scheme for Krypton Polar, including ACK cluster version upgrade, cloud native gateway replacement, operating system upgrade, etc. Migrate the microservice gateway, traffic gateway, Nginx ingress, API gateway, etc. to the cloud native gateway, reduce unnecessary link consumption and maintenance costs of different products, improve the observability of core links and reduce network latency. Based on Alibaba Cloud ARMS products, observable standards are formulated for different roles' concerns in development, testing, and operation and maintenance, covering different scenarios such as front-end, APP, microservices, and containers, so as to achieve end-to-end full stack performance monitoring and end-to-end link tracking capabilities, and improve the ability to locate problems or failures.

In order to clarify the concurrent capacity that the Polar Krypton APP can actually carry, and assist Polar Krypton in multi round full link pressure testing and tuning of the core business interface. Through full link pressure testing, not only has the effect of special transformation schemes such as grayscale release, lossless online and offline, and elastic scalability been verified, but also its confidence in switching the production system to the cloud native architecture has been enhanced.